CSIRO PUBLISHING
www.publish.csiro.au/journals/wr
Wildlife Research, 2006, 33, 615–626
Powerline corridors: degraded ecosystems or wildlife havens? Donna J. ClarkeA,B, Kate A. PearceA and John G. WhiteA A
School of Life and Environmental Sciences, Burwood Campus, Deakin University, 221 Burwood Highway, Burwood, Vic. 3125, Australia. B Corresponding author. Email:
[email protected]
Abstract. Management of powerline corridors in Australia has traditionally focused on the complete removal of vegetation using short rotation times owing to the perceived hazard of fire associated with corridor vegetation. Because of the intense management associated with fire hazards, little thought has been given to use of powerline corridors by wildlife. This has resulted in corridors traditionally being viewed as a source of fragmentation and habitat loss within forested ecosystems. We investigated the responses of small mammal communities living in a powerline corridor to management-induced vegetation changes at different successional stages, to determine whether a compromise could be reached between managing corridors for fire and biodiversity. Habitat modelling in the corridor and adjacent forest for three native and one introduced small mammal species demonstrated that species responded to changes in vegetation structural complexity, rather than time-since-management per se. Early seral stages of vegetation recovery after corridor management encouraged the introduced house mouse (Mus domesticus) into corridors and contributed little to biodiversity. Mid-seral-stage vegetation, however, provided habitat for native species that were rare in adjacent forest habitats. As the structural complexity of the vegetation increased, the small mammal community became similar to that of the forest so that corridor vegetation contributed fewer biodiversity benefits while posing an unacceptable fire risk. If ecologically sensitive management regimes are implemented to encourage mid-seral vegetation and avoid complete vegetation removal, powerline corridors have the potential to improve biodiversity. This would maintain landscape connectivity and provide habitat for native species uncommon in the forest while still limiting fuel loads in the corridor. Introduction Powerline corridors and rights-of-way are generally seen as a source of habitat fragmentation and loss for many forestdependent species, and are perceived to hold little biodiversity value. The fragmentation aspects of powerline corridors for forest-dependent species has been discussed extensively in the literature (e.g. Hanowski et al. 1993; Goldingay and Whelan 1997; Macreadie et al. 1998; Goosem 2000), but little research has focused on the potential biodiversity benefits of corridor vegetation in forested ecosystems. Powerline corridor management has traditionally focused on the reduction of woody vegetation and short rotation times between management events, with little consideration given to any potential use of corridor vegetation by wildlife. Current legislation in Australia states that vegetation must be a minimum of 6.75 m from the lowest point of the cables at maximum sag. Based on this, vegetation can grow 4.75 m in height before management must be implemented on 500-kV lines. However, fire hazards, concern over system reliability and a general lack of information on environmental impacts has led to clearance of most vegetation from underneath powerlines. Traditional techniques focus on complete vegetation removal on rotations of 1–6 years depending on habitat © CSIRO 2006
type, via the use of broad-acre spraying with dicot-specific herbicides, and handcutting and/or mowing (Arner et al. 1976), resulting in a community dominated by short grasses (Bramble and Byrnes 1976; Carvell 1976; Bramble et al. 1990). These management techniques often result in the creation of hard edges by forming a high-contrast early-successional vegetation community, which is lower and less structurally complex than the surrounding matrix vegetation (Forman 1995). This causes an increase in associated edge effects, decreased landscape permeability for interior forest species and increased permeability for ubiquitous, generalist species (Forman 1995). This change in landscape characteristics frequently creates barriers to movement for some species and causes loss in landscape connectivity. Ultimately, vegetation within powerline corridors is characterised by a loss of species diversity (Carvell 1976; Brown 1995) and is seen to offer limited functional habitat for biodiversity (Brown 1995). Powerline corridor management within the USA and Canada, however, is moving towards ecologically sensitive vegetation-management techniques that enhance potential biodiversity benefits. Considerable research in these countries has focused on the effects of management on food and browse resources for game species and other wildlife 10.1071/WR05085
1035-3712/06/080615
616
Wildlife Research
(Bramble and Byrnes 1972; Arner et al. 1976; Doucet et al. 1983). In contrast, corridor management in Australia is still focused on complete vegetation removal because vegetation, and consequent fuel loads, is perceived to pose a risk to the structure, function and integrity of transmission lines. This strong focus on fuel reduction has led to limited research within Australia into the ecological changes resulting from the long-term management of forest clearings for powerline corridors (Goldingay and Whelan 1997; Goosem and Marsh 1997; Baker et al. 1998). Although Australian power companies have a responsibility to manage corridor vegetation to maintain a safe, secure supply of electricity, studies in the USA (Schreiber and Graves 1977; Johnson et al. 1979) and in Australia (Goosem and Marsh 1997; Macreadie et al. 1998) have found that modifying the powerline corridor habitat can provide habitat for several species not commonly found in the surrounding forest. This paper aims to investigate the responses of small mammal communities living in corridor vegetation to management-induced vegetation changes. Corridor vegetation managed on different time scales and therefore at different successional stages was surveyed to assess the effects of successional stage on fuel loads and small mammal communities to determine: (1) whether powerline corridors can offer benefits to biodiversity, and (2) whether a balance can be reached between managing powerlines for fire hazard and providing habitat for wildlife. Materials and methods
D. J. Clarke et al.
categories (3+ years) owing to the general lack of areas within the corridor that had been left that long since management. A trapping grid consisting of 40 aluminium folding traps (10 × 10 × 33 cm) (Elliott Scientific, Upwey, Victoria) was set at each site. Traps were placed 25 m apart in eight parallel transects, spaced at 20-m intervals. Four transects were set in the powerline corridor, and two transects were set in each side of the adjacent forested areas parallel to the corridor. Traps were baited with a mixture of peanut butter, rolled oats, honey, tuna oil and linseed oil. Traps were set for five consecutive nights and cleared each morning. Captured animals were identified to species, and reproductive condition, head–body measurements and location were recorded. All individuals were ear-notched to identify previous captures and were released at the point of capture. Floristics and structure of vegetation Vegetation complexity, diversity and life form were surveyed at all 25 sites at five random positions in a quadrat (5 m × 5 m) around each trap location. Vegetation complexity was measured using a 2-m pole divided into 10-cm increments, placed vertically at ground level with a maximum of 10 contacts recorded for each increment. Estimates of plant abundance were undertaken to measure floristic composition of sites. Percentage cover of Leptospermum sp., Hakea sp., Bankisa sp., Acacia sp., Xanthorrhea sp., and Melaleuca sp. were estimated. Sedges (Gahnia sp., Lomandra sp., Carex sp., Juncus sp. and Restio sp.), ferns (Blechnum sp., Gleichnia sp., and Lindsaea sp.), and introduced species (Onopordum sp., Rubus sp., Urtica sp., and Senecio sp.) were also classified into separate floristic groups. Vegetation surveys additionally recorded the percentage cover of life forms, leaf litter, rock, and bare ground and the number of logs around each trap location. Life forms measured were canopy cover, native grass, introduced grass, tall and short shrubs, tall and short herbs, moss and vines. All vegetation surveys were conducted at the time of trapping to ensure that species abundance and responses were related to the current vegetation characteristics.
Study area Twenty-five sites were established along a 100-m-wide powerline corridor bisecting Bunyip State Park and the adjoining Kurth Kiln Regional Park in Victoria, Australia. These parks are located 65 km south-east of Melbourne and cover 20000 ha. The double 500-kV transmission line was established in 1962 and crosses the north-east of Victoria from Hazelwood in the north to South Morang, where power is distributed to stations around Melbourne (Parks Victoria 1998). The powerline corridor extends ~20 km through the northern section of Bunyip State Park, and through the centre of Kurth Kiln Regional Park (Parks Victoria 1998). SPI PowerNet and Parks Victoria currently jointly manage the Bunyip State Park and Kurth Kiln Regional Park sections of the corridor. Vegetation-management techniques used to minimise risk to the transmission lines during wildfire have significantly modified the structure and composition of the seven vegetation communities crossed by the powerline. The use of broadcast spraying, spot spraying, slashing, blading and clearing within the corridor, generally on a 3-year cycle, has led to the removal of canopy cover and reduction of shrub and ground layers (Parks Victoria 1998). The presence of a 5-m-wide utility road, absence of fire, and invasion by pest plant species has also led to a general degradation of the quality of the vegetation communities. Small mammal surveys Live trapping for small mammals was undertaken once at 25 sites between June and December 2002. All sites were at least 500 m apart. On the basis of management-history records, sites were grouped into three time-since-management categories: 1–3 years (16 sites), 3–5 years (5 sites) and >5 years (4 sites). There were few sites in the latter
Fuel hazard surveys Overall fuel hazard was assessed at each trap location, using surface fuel, elevated fuel and bark fuel measurements, as described in McCarthy et al. (1999): Overall fuel hazard = (sum of the influences of) bark hazard + elevated fuel hazard + surface fine fuel hazard Surface fuel hazard was recorded using a fuel depth gauge to measure depth of leaf litter at five random points around each trap location. These values were averaged to produce a fuel hazard rating of low, moderate, high, very high or extreme (McCarthy et al. 1999). Elevated fuel was assessed by recording the height of vegetation, amount of dead material and density of foliage and twigs within the ground to mid-storey strata. Bark fuel was estimated by recording the attributes of the upper-storey species present. The type of bark and how loosely the bark was held to tree species were recorded. Both the elevated fuel and bark hazard were visually estimated, on the basis of descriptions in McCarthy et al. (1999). All three fuel hazard ratings for each trap location were converted to an average equivalent fuel load (t ha–1), as described in McCarthy et al. (1999). Average fuel load was estimated for both the forest and corridor habitats, for all sites. Statistical analyses Habitat characterisation All vegetation variables were averaged across the forest and the powerline corridor at each site separately, to give a proportion for each variable for both habitat types. Principal component analysis used VARIMAX rotation to reduce multicolinearity between the structure variables only, and produced factors with eigenvalues over 1.0 to reduce
Powerline corridor management and small mammals
the number of structure parameters in the models. A non-parametric Wilcoxon signed-ranks test assessed the differences in habitat composition, species richness and fuel loads between the forest and adjacent corridor habitats. SPSS for Windows, Rel. 10.0.0. (SPSS Inc., Chicago, 1999) was used to conduct both analyses. Species abundance models The most common species were modelled separately using R statistical packages (Ihaka and Gentleman 1996). Prior to analysis, an initial exploratory approach investigated the relationship between species abundance (defined as individuals known to be alive: KTBA) and the habitat variables measured during vegetation surveys. Models for Mus domesticus (house mouse) used the number of captures as the abundance measure as insufficient individuals were captured to run the models. The inclusion of habitat variables in models was selected primarily on a priori knowledge gained from extensive literature reviews of the most abundant species across the study. Different variables were selected for each species on the basis of literature evidence suggesting which habitat variables should be important for a species. Following a priori selection, hierarchical partitioning using the hier.part package in R (Walsh and Mac Nally 2003), was used to select five predictor variables with a high independent influence on species abundance to be included in the models. The data were modelled using the information-theoretical approach described by Burnham and Anderson (2002). The Akaike information criterion corrected for small sample sizes (ratio of sample size to the number of parameters was 0.9) (Anderson et al. 2001; Burnham and Anderson 2002). Weighted-model averaging based on 5000 bootstrapped samples was applied across all models to reduce model selection bias. Algorithms were used to calculate AICc, QAICc, bootstrap frequencies, and modelaveraged estimates in R (M. Scroggie, unpublished).
Wildlife Research
Habitat characterisation Principal component analysis of vegetation structural complexity data produced two clear factors, structure from 80 to 200 cm and from 0 to 80 cm, for use in the species abundance models, accounting for 95.6% of structural variation. Continual management of the corridor to reduce fuel loads has led to changes in the composition of the community within the corridor. The corridor had higher mean values of bare ground, short herbs, sedge sp., Acacia sp. and Leptospermum sp. when compared with the forest (Table 1). In contrast, tree-based resources such as canopy cover, leaf litter and logs characterised the forest habitat. The forest community also showed higher mean values of ferns, midstorey shrubs and trees such as Banksia sp., Hakea sp. and grass tree species (Xanthorrhoea sp.), when compared with the corridor community (Table 1). Small mammals Seven native and one introduced small mammal species were trapped during 5000 trap-nights. The most abundant species trapped included Rattus fuscipes, Rattus lutreolus, Antechinus agilis and Mus domesticus. The dusky antechinus (Antechinus swainsonii), eastern pygmy possum (Cercartetus nanus), long-nosed bandicoot (Parameles nasuta) and white-footed dunnart (Sminthopsis leucopus) were caught at much lower frequencies, and were omitted from further analysis. All species except C. nanus were trapped at least once in the corridor habitat. M. domesticus, S. leucopus and P. nasuta were not trapped in the adjacent forest (Table 2). There was no significant difference in species richness between the corridor and forest (Z = –0.587, P > 0.05). Small mammal community composition and abundance, however, were different between the forest and the corridor habitat. M. domesticus (Z = –2.388, P < 0.05) and R. lutreolus (Z = –2.294, P < 0.05) responded positively to the presence of the Table 1. Differences in the mean percentage cover of habitat variables in the corridor versus the forest shown by the Wilcoxon signed rank test Significant differences only are shown Mean percentage Forest Corridor
Time-since-management models Changes in habitat characteristics with time-since-management were modelled for the corridor data only, using generalised linear models assuming a Poisson distribution. Weighted-model averaging and 5000 bootstrapped samples were used to make inferences about the best model.
Results High species richness and abundance of small mammals were found within corridor habitat and forest habitat across the study. All species except one were trapped at least once in the corridor habitat.
617
Canopy cover Bare ground Leaf litter Logs Short herbs Sedges Ferns Banksia sp. Acacia sp. Hakea sp. Leptospermum sp. Xanthorrhea sp.
42.83 9.00 60.43 28.60 9.31 21.84 23.94 7.36 7.70 4.34 11.01 1.77
1.76 20.26 0.09 5.36 12.92 31.76 18.58 0.46 10.67 1.53 16.81 0.00
Z
P
–4.356 –3.673 –4.373 –4.373 –2.600 –4.038 –2.658 –3.533 –2.277 –2.848 –3.338 –2.041
